HP-UX

system is HP-UX B.11.23 U ia64

A standard feature of almost all of our cron jobs is to redirect output to a log file, then at the end of the script to embed the contents of that log file in the body of an email sent to the IT staff. Typical code to do this is

Code:

$MAILER -s "$PROC: $SUBJECT" $MAILTO < $LOG > /dev/null 2>&1

where $MAILER = /usr/bin/mailx

the mail is forwarded to an Exchange server where we then pick it up in our individual Outlook accounts. There are several dozen, possibly in excess of 100 of these every day. Last week we quit receiving the email from one single job. All others continue to come through. I checked for the log file output of the job (set to /tmp/$PROC.$$) and it is being created and shows the job ran as expected. But the mail appears to not be reaching our Exchange server. Or if it is, it is not making it to our individual accounts. On the source, the mail is sent to a distribution list defined in our aliases file.

So, how do I begin to follow this email through the system and find where it is getting dropped? (My primary job role is Oracle DBA, but I work closely with our unix SA and consider myself a power user or SA apprentice. I have full root access to the HP system. The exchange server is managed by a different dept and I have no direct access to it, will have to work with their admin when we get to that point.

I would start by:
What has this job so special compared to the others? For if all others make it there must be something that makes the difference...
try redirect STDOUT to something like $HOME/cronjobID_mailer.log and and 2 >jobID_mailer.err rather than

Code:

 > /dev/null 2>&1

And see what output you get...

Thanks for the reply. Nothing about the job itself differs from any other. And this one *was* sending mail just fine until 22 Sep. We had changed nothing in the script itself or the mail config when it stopped. Further testing indicates there is an issue with the output of the key program being executed by the job (oracle's rman backup utility). That output is included in the body of the email and by doing some focused testing - controlling exactly what actions I have rman report on - the mail will go through. Narrowing that down now.

Have you tried to separate and put in file the redirections?
I would add an extra mail line for test at that precise point subject: after mail send... to yourself and see if you get it
Ive seen cases where in logs, we get side effects because of some characters I suppose, where the input never ends... (waiting for line feed or dot or...) but I believe such case should show up somewhere... I never managed to recreate the situation (without getting myself in a mess...)
So I suppose you could try to write a script that loads nothing and see what you get...
Does the log exist? I there anything in /var/spool/cron/log?